A model of katsushika hokusai (oui), trained only with images of his second ascension outfit.

The clothing of this character is very complex, so I decided to train this Lora with specific images only, because by doing it with images of both kimonos, and using the prompt "kimono", these get mixed up and don't generate the clothing it is supposed to be.

The results? Well, it's not perfect, but it's pretty close after a few tries.